Electronic and Geometric Features of (q5-C5H5 ... - ACS Publications

Organometallics 1983,2, 1273-1284 5

+ P h a P was determined to be 1 X

M. Preparation of the Methoxy Analogue of 4. To a solution of 167 mg (0.3 mmol) of 3 in 100 mL of CHC13 was added 100 mL of methanol with magnetic stirring. Upon the subsequent addition of 384 mg (1.47 mmol) of Ph3P, the solution immediately became darker and then lightened in color over a 1-2-h period. Ten hours after the addition of Ph3P, the solution was rotoevaporated to a yellow oil that when triturated with pentane, rotovapped, and dried under vacuum gave a yellow-white solid. The solid was extracted with 20 mL of E t 2 0 to remove unreacted Ph,P and residual methanol. After rotoevaporation and drying under vacuum, 256 mg of a white solid was obtained. A portion of this was dissolved in CDCl,, and sufficient Ph,P was added to cause an upfield shift in the vinyl proton resonances which avoided obscuring the CHCl proton resonance of 4. Integration of the CHCl resonance of 4 vs. the methoxy analogue resonance gave a 85:15 molar ratio of methoxy analogue to 4. Following the combination of this NMR tube contents with the remainder of the white solid, the material was chromatographed on silica with CHC1,. An early eluting fraction was determined by NMR spectroscopy to contain a mixture of 4 and methoxy analogue, while the major and later eluting fraction, following rotoevaporation, pentane trituration, and drying under vacuum, yielded 118 mg (46%) of the white methoxy analogue; mp 205 "C. Anal. Calcd for C4,H4,0P2Pt: C, 60.59; H, 4.86. Found: C, 60.37; H, 4.92. Control experiments were run deleting either Ph3P or MeOH resulting in no significant reaction according t o NMR spectroscopy. In addition, solutions of 4 or 5 that were incubated in MeOH(CHC1,) showed no production of the methoxy analogues. X-ray S t r u c t u r a l Determination of 4. Crystals of 4 were grown a t the interface of a heptane/chloroform bilayer and a specimen suitable for X-ray analysis was mounted on a fiber. X-ray data collection was carried out on a Nicolet R3m automated diffractometer equipped with a Mo target X-ray tube (A = 0.71073 A) and a graphite crystal monochromator.8 Unit cell constants were determined to be a = 12.804 (5) A, b = 13.061 (5) A, c = 13.416 (5) A, CY = 65.95 (3)", @ = 67.59 (3)', and y = 8.45 (3)' for a cell of triclinic symmetry. The space group was determined to (8) Programs used for centering of reflections, autoindexing, refinement of cell parameters, axial photographs, and data collection were those described in: 'Nicolet P3/R3 Data Collection Manual"; Calabrese, J. C., Ed.; Nicolet XRD Corp.: Fremont, CA, 1980.

1273

be Pi. Intensity measurements of 3669 independent reflections for 3 I 20 I45" were observed with Z 5 3 a Q . Absorption corrections were made by using the empirical psi scan method, and no decay corrections were necessary. The structure was solved by direct methods that revealed the locations of all non-hydrogen atoms on the initial E map. The structure was refined to a final R value of 0.065 by a blockedcascade technique with anisotropic thermal parameters for all non-hydrogen atoms. Hydrogen atoms were placed in idealized positions with isotropic thermal parameters. All structural determinations and refinement calculations were carried out with the SHELXTL package on the Nicolet R3m crystallographic ~ y s t e r n . ~A cell mass of 1793.5 containing 892 electrons and two molecules per unit cell gave a calculated density of 1.60 g/cm3 for CMH&C12P2. The final difference map revealed no abnormal features.

Acknowledgment. We express o u r g r a t i t u d e for financial support of t h i s research b y Montana State University and the National Science Foundation, G r a n t CHE 7826160. We also gratefully acknowledge work of C h u c k Campana and t h e use of an R3m Crystallographic System provided b y the Nicolet XRD Corp. Finally we wish to thank Dr. James F r y e and the Colorado State University Regional NMR C e n t e r , f u n d e d t h e N a t i o n a l Science Foundation ( G r a n t CHE-78-18581) for o b t a i n i n g the MAS/CP NMR spectra. Registry No. 1, 3635-94-7; 2, 86834-41-5; 3, 86823-57-6; 4, 86823-587; 4 (MeO-deriv.),86823-60-1; 5,86823-59-8; (C2H4PtC12),, 12073-36-8; Ph,P, 603-35-0. Supplementary Material Available: Tables of bond lengths, bond angles, atom coordinates and temperature factors, anisotropic temperature factors, and hydrogen coordinates and temperature factors (5 pages). Ordering information is given on any current masthead page. (9) Programs used for data reduction, Fourier syntheses, direct method structure solution, least-squares refinement, error analysis, leastsquares planes calculation, and calculation of hydrogen position are those described in: "Nicolet SHELXTL Structure Determination Manual"; Sheldreck, G. M., Ed.; Nicolet XRD Corp.: Fremont, CA, 1980. (10) Cheng, A. K.; Stothers, J. B. Org. Magn. Reson. 1977, 9, 355.

Electronic and Geometric Features of (q5-C5H5)ML16-Electron Fragments. A Molecular Orbital Study of Ligand Effects Peter Hofmann * and Moothetty Padmanabhan Institut fur Organische Chemie der Universitat Erlangen, 0-8520 Erlangen, Federal Republic of Germany Received March 9, 1983

Qualitative M O theory, based upon extended Huckel calculations, is utilized t o study coordinatively unsaturated 16-electron fragments d8 (q5-CSH5)ML (M = transition metal, L = ligand) with respect t o their geometric and electronic structure as a function of the ligands L. It is shown t h a t such fragments, observed as reactive intermediates in organometallic reactions, can possess singlet ground states, which then will exhibit "nonlinear" Cp-M-L arrangements. T h e comparative ease of a number of important reactions proceeding via such intermediates without a n y spin-imposed high activation barriers can thus be readily understood.

Introduction Since the early and i m p o r t a n t kinetic e x p e r i m e n t s b y Bas010 and Schuster-Woldan,' i t has been well-known that ds 18-electron half-sandwich c o m p o u n d s of the t y p e (71,C5H5)M(C0)2 (1, M = Co, Rh, Ir) undergo t h e r m a l ligand To whom correspondence should be addressed at the Anorganisch-chemisches Institut der Technischen Universitat Miinchen, D-8046 Garching, FRG.

0276-7333/83/2302-1273$01.50/0

Scheme I

-

1 -

0

2

-

3

Substitution reactions b y associative ( S N ~ )pathways.* T h i s p u t s t h e m i n c o n t r a s t to m e t a l carbonyl complexes

0 1983 American Chemical Society

-

- -

Hofmann and Padmanabhan

1274 Organometallics, Vol. 2, No. 10, 1983

Scheme I11

Scheme I1

I cp

it 0 ’ 0

1

hv

-co

l

co

l

-. . . .

C

0

-4

like Ni(C0)4,Fe(C0)5,Cr(CO)6,etc., which typically react by CO loss3 via “unsaturated” (more or less solvent stabilized) 16-electron intermediates, as exemplified by Cr(c0),.4The mechanism a t present believed to be operative for dicarbonyl species CpM(CO), (1, Cp = v5-C5H5) is shown in Scheme I. Any d8 CpM(C0) intermediates do not seem to be involved. The accessibility of (1,C6H5)M(C0),Ltransition states or intermediates 2, formed by nucleophilic attack of L at the metal, avoids unfavorable 20-electron configurations through a simple “slipping” motion of the ML, fragment., Indeed on the basis of further experimental evidence6 a general rule has been suggested by Basolo recently,’ saying that associative substitution mechanisms will play a role whenever the metal can localize a pair of electrons onto an appropriate ligand,8 as shown for Cp in 2. Contrastingly, there have been several reports in the recent literature that convincingly indicate a thermally or photochemically induced formation of 16-electron intermediates of the d8 CpML type. Years ago Brintzinger9had postulated the formation of an intermediate CpCo(C0) (4) in solution photochemistry experiments on the parent dicarbonyl CpCo(CO), (Scheme 11). The monocarbony14 was reported to undergo fast transformations to a number of binuclear and trinuclear complexes. A more recent reinvestigationlo of the photochemistry of 1,although in inert-gas matrices (Ar, N,), seems to indicate, however, that no CpCo(C0) (4) is formed a t all and that Brintzinger’s intermediate may really have been the dinitrogen complex CpCo(C0)(N,) (3, L = N,). Moreover, these matrix isolation studies revealed that in a CO matrix a species is reversibly formed by irradiation, the vibrational spectra of which can be plausibly interpreted by assuming the presence of a Co(CO), subunit, consistent with an association of carbon monoxide to 1,yielding (v3-C5H5)Co(CO),(2, L = CO) in agreement with Basolo’s “rule”. Interestingly, the recent past has brought even more experimental observations, which clearly demonstrate that ds CpML fragments with 16 valence electrons do indeed (1) Schuster-Woldan, H . 4 . ; Basolo, F. J.Am. Chem. SOC.1966, 88, 1657. (2) For a recent review see: Basolo, F. Coord. Chem. Rev. 1982,43,7. (3) Werner, H. Angew. Chem. 1968, 80,1017; Angew. Chem., Int. Ed. Engl. 1968, 7 , 930. (4) Electronic structure calculations: Hay, P. J. J. Am. Chem. SOC. 1978, 100, 2411 and references therein. (5) For a general analysis of ML, shift processes (haptotropic shifts)

in complexes see: Albright, T. A.; Hofmann, P.; Hoffmann, R.; Lillya, C. P.; Doboah, P. A. J. Am. Chem. SOC.1983, 105, 3396. Anh, N. T.; Hoffmann, R. Ibid. 1978, 100, 110. (6) Chang, C. Y.; Johnson, C. E.; Richmond, T. G.; Chen, Y. T.; Trogler, w. C.; Basolo, F. Inorg. Chem. 1981, 20, 3167. (7) Basolo, F. Inorg. Chim. Acta 1981,50, 65. For a theoretical generalization see: Chu, s.-Y.;Hoffmann, R. J. Phys. Chem. 1982,86,1289. ( 8 ) Cp2W(CO)2,a formal 20-electron complex, exhibits this localization in its ground-state structure with one Cp ring being v3 bound: Huttner, G.;Brintzinger, H. H.; Bell, L. G.; Friedrich, P.; Bejenke, V.; Neugebauer, D. J. Organomet. Chem. 1978, 145, 329. Other cases of ground-state localization of a pair of electrons on a Cp ring include PdzLz(pCp)ol-X), Pd2L2( ~ - C P )or~ CpPdR(PR)3 , molecules: Werner, H. Adu. Organomet. Chem. 1981,19,155. Werner, H.; Kraus, H.-J.; Schubert, U.; Ackermann, K.; Hofmann, P. J. Organomet. Chem. 1983,250, 517. (9) Lee, W.-S.; Brintzinger, H. H. J. Organomet. Chem. 1977,127,87. (10) Crichton, 0.; Rest, A. J.; Taylor, D. J. J. Chem. SOC.,Dalton Trans. 1980, 167.

-5

-6

-7

Scheme IV

9 =

io;

10 -

exist as reactive intermediates. For the case of CpIr(CO)2, Rausch” has postulated the formation of CpIr(C0) as the active species formed on photolysis of the dicarbonyl. Similarly Werner and Leonhardl, have pointed out the intermediacy of a related 16-electron fragment, CpCo(PMe,), generated via PMe, abstraction from CpCo(PMe,), by the “masked” Lewis acid CpMn(CO),(THF). In addition their experiments indicate the generation of CpCo(C0) from CpCo(CO), in an analogous fashion. Cp*Ir(CO) (Cp* = q5-C5Me5)was found by Graham et recently in photochemical studies of Cp*Ir(CO)2and was shown to be capable of oxidatively adding to CH bonds of saturated hydrocarbons. Bergman and Janowicz14 generated the unsaturated fragment Cp*Ir(PPh3) from Cp*Ir(PPh3)H2by light-induced H, loss; again this species is able to insert into CH bonds of hydrocarbons. In a similar study, Jones and Feherl, claimed to have obtained Cp*Rh(PMe,) as an intermediate product from Cp*Rh(PMe,) (ary1)H molecules both by thermal and photochemical reactions. Even more fascinating and intriguing is the observation by Shapley and co-workers,16 that CpIr(CO)H, is clearly in thermal equilibrium with H, and CpIr(CO), which can be trapped by appropriate ligands or causes the formation of Cp,Ir3(CO)3. We became especially interested in this context by yet another paper from the Bergman proup,l7which essentially formed the impetus for this study. Prompted by a prethat ligand dissociations in d8 CpML, diction of C~llman,’~ complexes might be spin forbidden, e.g., should possess abnormally high barriers (thus offering a rationale for preferred associative pathways), these authors carried out a kinetic study of phosphine substitution in CpCo(PPh,), (5). The objective was to test the validity of the assumption that intermediates of the CpML type should be forced by symmetry to have a triplet ground state for a d8 electron count and should therefore not be easily accessible as intermediates in thermal organometallic reactions. Similar arguments of a spin-induced high barrier (11) Rausch, M. D.; Gastinger, R. G.; Gardener, S. A.; Brown, R. K.; Wood, J. S. J. Am. Chem. SOC.1977, 99, 7870. (12) Leonhard, K.; Werner, H. Angew Chem. 1977,89,656; Angew. Chem., Znt. Ed. Engl. 1977, 16, 649. (13) Hoyono, J. K.; Graham, W. A. G. J. Am. Chem. SOC.1982,104, 3723. (14) Janowicz, A. H.; Bergman, R. G. J.Am. Chem. SOC.1982,104,352. (15) Jones, W. D.; Feher, F. J. J. Am. Chem. SOC.1982, 104, 4240. (16) Shapley, J. R.; Adair, P. C.; Lawson, R. J.; Pierpoint, C. G. Inorg. Chem. _ . 1982.21. 1701. (17) Janowicz, A. H.; Bryndza, H. E.; Bergman, R. G. J. Am. Chem. SOC.1981,103, 1516. (18) See footnote 5 in ref 17. We are also grateful to Prof. Collman ~~~

I~

for bringing this prediction to our attention: Collman, J. P., private communication.

( $ - C a S ) M L 16-Electron Fragments have been forwarded by Collman and Hegeduslg for reactions involving d2 Cp2ML complexes, although experimental evidence for Cp2W-C2H5+is available.20 Cp2ML systems have been studied theoretically by Hoffmann and LauherZ1and for a d2 electron count might well deviate from CZvsymmetry, the assumption of which led to Collman’s proposal. For systems like the monocarbonyl 4, CBUsymmetry assumed, high-spin character is indeed strictly required as can be easily seen from the orbital pattern of a d8 CpM conical fragmentz2interacting with a carbonyl ligand (vide infra). Surprisingly, Bergman found that the substitution of PPh3 in 5 by PMe,, leading to the mixed bis(phosphine) system 7, not only rapidly takes place a t very low temperature but also definitely involves the formation of CpCo(PPh,) (6) in the rate-determining dissociative step (Scheme 111). Intermediate 6 is then irreversibly trapped by PMe, in a very fast second step. A final piece of evidence for CpML 16-electron intermediates to be cited here is found in Bergman’s investigation of an NO insertion reaction as shown in Sheme IV.23 From kinetic data it was concluded that 9, whatever its detailed structure may be, is easily formed from 8 in a rate-determining step and is then intercepted by added phosphine in a fast consecutive reaction yielding 10. With all the above cited evidence available, we thought it worthwhile to take a closer look a t the electronic structure of some d8 CpML model systems. It is unclear what geometry and which spin state is to be expected for such fragments, and it is the purpose of this paper to decipher some of the dominant molecular orbital features that play a role in determining their structures and their tendency toward singlet or triplet ground states. Incidentally, d8 CpML species probably also are important intermediates in the well-documented alkyne oligomerizations mediated for instance by C ~ C O ( C Oand )~~~ CPCO(PR~)~.~~ Obviously species like 4, “linear” CpCo(CO), CbU,in contrast to the 18-electron analogues C p N i ( N 0 P or CPCU(PR,),~ are bound to be Jahn-Teller unstable if they were to adopt a single ground state ?Ez).As already speculated by Bergman,17 this fact might lead to a preferred “bent” ground-state geometry for isoelectronic phosphine-containing fragments like 6 or like CpRh(PMe,), Cp*Ir(PPh3),etc., allowing these intermediates to stay on the singlet (low-spin) surface, thus circumventing any spin-forbiddenness problems for reactions passing through d8 CpML species. We have discussed the electronic structure of “bent” d8 CpM(C0) fragments as building blocks of metal-metal bonded systems else(19)Collman, J. P.;Hegedus, L. S. “Principles and Applications of

Organotransition Metal Chemistry”; University Science Books: Mill Valley, CA, 1980. (20)Hayes, J. C.;Pearson, G. D. N.; Cooper, N. J. J.Am. Chem. SOC. 1981,103,4648. (21)Lauher, J.; Hoffmann, R. J . Am. Chem. SOC.1976,98,1729 and

references cited therein. (22)Elian, M.; Chen, M. M. L.; Mingos, D. M. P.; Hoffmann, R. Inorg.

Chem. 1976,15,1148. (23)Weiner, W. P.; White, M. A.; Bergman, R. G. J. Am. Chem. SOC. 1981, 103,3612. (24)For a review see: Vollhardt, K. P. C. Acc. Chem. Res. 1977,10, 1 and references therein. Vollhardt, K. P. C.; Bergmann, R. G. J . Am. Chem. SOC.1974,96,4996. (25)Yamazaki, H.; Wakatsuki, Y. J. Organomet. Chem. 1977,139,157. Wakatsuki, Y.; Kuramitsu, T.; Yamazaki, H. Tetrahedron Lett. 1974, 4549. McAlister, D. R.;Bercaw, J. E.; Bergman, R. G. J. Am. Chem. SOC. 1977,99,1666. (26)Fischer, E. 0.; Jira, R. 2.Naturforsch., B: Anorg. Chem., Org. Chem., Biochem., Biophys., Biol. 1954,98,518.Cox, A. P.;Thomas, L. F.; Sheridan, J. Nature (London) 1958, 181,1157. (27)Cotton, F. A.;Takata, J. J. Am. Chem. SOC.1970,92,2353.

Organometallics, Vol. 2, No. 10, 1983 1275

t -7

t

EleVl

2

,,

“sp”

I

-11

-12

-13

-14

t0

L L

Figure 1. Orbital interaction diagram for a model system CpCoL (L = u donor, “linear” C , geometry). Only the metal contributions

to the wave functions of CpCo are shown. The numbering of the symmetry labels of MO levels corresponds to only the valence levels here and in all following cases.

where.28 They are isoelectronic to Fe(CO),, for which a triplet ground state was found in matrix isolation studies.% Here we make use of extended Huckel calculation^^^ to inspect more closely the orbital structure of various de CpML systems as a function of L and its electronic properties. Although our calculations using the EH methodology will of course not allow us to determine absolute energies for singlet vs. triplet states,31they should provide guidelines for a general qualitative analysis of geometric and electronic features in CpML species. In the following section we discuss the electronic structures of de CpML fragments for different prototypes of ligands L.

Results and Discussion de CpML (L = o Donor). Let us start with ligands having rotational symmetry (linear, monoatomic, C3”,etc.) and here with the simplest case of a CpML system, where (28)Hofmann, P. Angew. Chem. 1979,91,591; Angew. Chem., Int. Ed. Engl. 1979, 18,554.Pinhas, A. R.;Albright, T. A,; Hofmann, P.; Hoffmann, R. Helu. Chim. Acta 1980,63,29. (29)Poliakoff, M.; Turner, J. J. J . Chem. SOC.,Dalton Trans. 1974, 2276. Barton, T. J.; Grinter, R.; Thomson, A. J.; Davies, B.; Poliakoff, M. J. Chem. SOC.,Chem. Commun. 1977,841.For MO-based predictions see: Burdett, J. K. J. Chem. SOC.,Faraday Trans. 2 1974, 70,1599.The

molecule is of Czysymmetry. (30)Hoffmann, R. J. Chem. Phys. 1963,39, 1397. Hoffmann, R.; Lipscomb, W. N. Ibid. 1962,36,3179,3289; 1962,37,2872. (31)The reader should r e d here the long history and difficulties (and costs) encountered in quantitatiuely reliable calculations of singlet-triplet splittings even for simple organic species like carbenes or carbene analogues. For a practically complete summary of semiempirical and ab initio work on carbenes see for example: Davidson, E. R. In ‘Diradicals”; Borden, W. T., Ed.; Wiley: New York 1982;Chapter 2.



change for a closed-shell configuration as a function of the bending angle a. In the reduced symmetry of 12 (C,)both former e sets (le, 2e) split as a decreases from 180’. While yz of 2e (28’’ in Figure 2) is unaffected by the bending motion (L moves on the nodal plane), x z (3a’) is destabilized because the antibonding interaction with the donor level of L is now turned on. The lifting of degeneracy brings out a noticeable HOMO-LUMO gap for the minimum energy structure of the low-spin case ( y z doubly occupied). The HOMO 28’’ is pure yz of the CpCo fragment while the LUMO 3a’ is mainly composed of x z , 2al, and the s-type orbital of L as schematically indicated in 13. 13, the u acceptor level of bent CpML, is nicely hy-

xz

7

7

I

I

180

160

- -

30 ’

13 -

-E

Le

I

201

51

I

I

140

,

I I

Figure 2. Walsh diagram and total energy variation for the transformation 11 12 (C5u C8). The bending here and in all other cases (unless otherwise noted) is in a mirror of the Cp ring. The asymmetry of total energy curves or Walsh diagrams induced by bending either toward the unique carbon of Cp or toward a C-C bond on the opposite side is only minute, in accordance with

a minute rotational barrier.

bridized toward the missing ligand of a parent CpMLz molecule. The dominant energetic stabilization upon bending comes from lal ( z 2 ) of CpCo, which had been interacting with the donor function of L in CBuin an antibonding fashion as shown in Figure 1. The donor orbital of L moves toward the nodal cone of z2 upon bending, and the strong antibonding interaction is reduced. In addition x z can now mix into this level in a bonding way from above, while the interaction of L with the well-hybridized 2al of CpCo is not perturbed very much. The ,$-derived level 2a’ therefore drastically drops in energy, finally undergoing an avoided crossing (see Figure 2) with la’ ( x 2 y2),which is increasingly destabilized by bending, because it feels more and more antibonding from L. The changes in overlap on bending for z2 and x 2 - y2 with s of L are indicated in 14 and 15; la’’ (xy) of course stays put for the same reason as 2a” ( y z ) .

L is a pure u donor. We can model such a group by a hydride-like pseudoligand L carrying a single 1s-type function occupied by two electrons. With Co as the metal center, d8 CpML (11) is a model for phosphine systems like CpCo(PPh3) (6) or CpCo(PMe3). We can disregard for the moment any potential acceptor capability of real phosphine ligands, which, however, will be added to our analysis later. The orbital structure of C, “linear” CpCoL is shown in Figure 1,the interaction diagram between a conical d8 CpCo fragment and L (L = H-). The valence orbitals of C,, CpCo have been discussed in the literature22and need 14 15 not be reiterated here. The s-type donor orbital of L interacts only with 2al, For our specific choice of L as H- with Hii = -13.6 eV, an empty hybrid at the metal center of CpCo with prelow-spin CpCoL is more stable in the bent geometry by . ~ ~ dominant sp character, and also with lal, mainly z ~ The about 12 kcal/mol, CY being about 137’. Of course for the latter is destabilized by the ligand orbital, and the anti18-electron case with two more electrons in the Walsh bonding combination of both levels mixes into itself 2al, diagram of Figure 2, a “linear” geometry will be preferred, in a bonding way toward L. Both e sets of CpCo ( l e ( x 2 and this is exactly what has been found for C ~ C U ( P R ~ ) . ~ ~ - y2,xy) and 2e (xz,yz)) cannot interact with L by symThe bent structure displays a HOMO-LUMO gap of 0.51 metry. Electronically a triplet ground state is mandatory eV between the formerly degenerate levels 3a’ and 2a” with for this geometry. An obvious escape from this situation xz and yz character. We cannot conclude from this numwould be possible by bending 11 to a structure shown in ber that any d8 CpML (L = CT donor) must be a bent singlet 12. Let us see what happens to the one-electron levels structure in its ground state, but we can say that by adopting or retaining a bent configuration (say after disI sociation of one group L from a CpMLp precursor), a dE co CpML fragment has a chance to stay on the singlet surface I L until it gets trapped. A symmetry-imposed high-spin L situation due to a linear structure can thus well be avoided. 11 12 This just might be the case for Bergman’s intermediate of Figure 1,if we go from 11 to 12. The results of a model CpCo(PPh3),thus explaining its ready formation by ligand calculation for our pseudoligand L are presented in the dissociation from C ~ C O ( P Pat~low ~ )temperature. ~ The Walsh diagram of Figure 2. Also shown is the total energy chance for a CpML system to remain on the singlet surface in a one-electron picture will depend on the actual size of the HOMO-LUMO gap of the bent structure and the gain (32) For simplicity we use the shorthand notation z2, x 2 - y 2 ,x y , x z , in total energy upon bending, both being a function of the yz, x , y , and z for dl*, d,zy*, ..., px, p,, and p1 orbitals, respectively.

-

-

- “L

( s 5 - C a 5 ) M L16-Electron Fragments


Table I. Geometric and Electronic Changes of o-Donor Svstems on Bending

CpML

deg

energy gainon

HOMO-

0.75 0.52 0.43 0.13 0.30 0.57 0.74

0.65 0.50 0.44 0.25 0.43 0.54 0.72

bending,b LUMO eV gap,eV

CPCOLC

& ( L ) = -12.0 eV

-13.6eV

-15.0 eV CpCo( PH,)d CpRh( PH,)d CpCo( CH,)CpRh(CH,)-

46.5 43.5 42.0 33.0 38.0 43.0

44.5

a Angle corresponding to minimum energy. For “singlet” model, Le., 2a“ doubly occupied. L is the model donor ligand with Is function, see text. No d orbitals on P.

180

160

1‘10

a

Figure 3. Total energy curves for bending of CpRhL (L = u donor): curve A, 2a“ doubly occupied; curve B, 2a” and 3a’ each singly occupied; dotted line, see text.

specific ligand L. The second determining factor, then, is the spin pairing energy of the metal, which, depending on the 3a’-2a” gap, may decide the system’s choice for a singlet or a triplet ground state. The Walsh diagram of Figure 2 is very reminiscent of the analogous situation for the organic “six-electron fragment” CH233and of carbenes in general, which are of course isolobal to a bent d8 Cph4La This may imply that, as in the case of CH2,a triplet system should also have a nonlinear (Cp-M-L) arrangement, though less bent than the singlet. Figure 3 shows the total energy change on bending as it derives from the Walsh diagram of Figure 2, if the two orbitals emerging from 2e (2a” and 3a’) are each occupied by one electron (curve B). Curve A, the total energy change for 2a” doubly occupied, analogous to Figure 2, is given for comparison. After the redistribution of the two relevant electrons, the bending angle a increases to 1 4 8 O , the one-electron level based total energy gain on bending is much smaller than for the occupation pattern. The values for Figure 3 are taken from calculations on CpRhL in order to exhibit the situation also for the heavier relative of cobalt, where an analogous picture exists. Both curves A and B, representing only the sum of EH orbital energy changes, of course, start out from the same energy for the linear Ck geometry, as our EHT model calculations do not take into account spin properties and mutual interactions of electrons. In reality the spin pairing energy will cause the triplet curve B to start out from a lower energy than curve A. If this shift is large enough to put the triplet minimum below that of the singlet (large spin pairing energy, dotted curve in Figure 3), then the ground state of a d8 CpML would be a bent triplet. It is clear that rigorous calculations, yielding reliable, fully geometry-optimized singlet and triplet energy values, comparable in quality to, say, state of the art CH2results, would involve an enormous computational effort and might even be not feasible at the present time. We resist the temptation here to derive any numeric values for state (33) See for example: Gimarc, B. M. ‘Molecular Structure and Bonding”: Academic Press: New York, 1979. (34) Hoffmann, R. Science (Washington, DC) 1981,211, 995. Hoffmann, R.Angew. Chem. 1982,94,725;Angew. Chem.,Int. Ed. Engl. 1982, 21, 711.

energies via the usual approach, which would involve spin pairing energies and electron interaction parameters for the ds metals in question here,35because we believe that such a ligand field type treatment superimposed upon EH calculations would not be very reliable either.36 From the well-known decrease of spin pairing energies within the Co, Rh, and Ir triad we can conclude, however, that the heavier metal atom systems d8 CpML (L = a donor as above) stand a better chance to exist as bent singlet species than their cobalt counterparts. Up to now we have made an arbitrary choice for our model donor ligand L, simply by using an ionization potential for the l s function of -13.6 eV. Let us analyze next, how a variation in the u donor character of L will influence our picture. We can, to a first approximation, mimic a better or worse u donor L by decreasing or increasing the orbital energy (Hiivalue) of L. The changes induced by such a variation, which we see for our purely a-interacting pseudoligand L, of course, carry over to other types of ligands, where the u donor capability is underlying any additional r donor or A acceptor interactions between M and L, which we will turn to later on. As can be seen from Figure 1, a higher lying donor orbital of L will lead to a more strongly destabilized 2a, level of CpML in the linear geometry, closer in energy to the half-filled 2e set. Bending of CpML for a better donor, therefore, should lead to a stronger stabilization of this level and in turn also to an increased destabilization of 3a’ ( x z ) , compared to the Walsh diagram in Figure 2. In terms of intramolecular perturbation theory3’ a better donor leads to a stronger mixing and repulsion of 3a’ ( x z ) and as soon as bending occurs. As a result, more energy 2a’ (9) will be gained on going from C, to C, and the gap between 3a’ and 2a” of the minimum structure of a bent singlet should increase with increasing donor capability of L. Opposite effects have to be expected for weaker u donors. Figure 4 shows the computed total energy curves and orbital energy changes for three different orbital energies of the Is function of L. The expected trends emerge from the model calculations: the best donor induces the highest energetic preference for a bent singlet structure (cf. Table I); 2a’ here experiences the largest stabilization and 3a’ gets (35) Slater, J. C. “Quantum Theory of Atomic Structures”;McCrawHill: New York, 1960; Vol. I. Williams, A. F. “A Theoretical Approach to Inorganic Chemistry”;Springer-Verlag: Berlin, 1979. Griffith, J. S. “The Theory of Transition Metal Ions”, Cambridge University Press: Cambridge, 1961. (36) In this connection see: Vanquinkenborne, L. G.; Haspeslagh, L. Inorg. Chem. 1982,21, 2448. (37) Libit, L.; Hoffmann, R. J. Am. Chem. SOC.1974, 96,1370 and references therein.



\

H,,(L)

=

-12.0

I

170

150

130

a

170

150

130

130

150

170

130 a

150

Figure 4. Total energy changes and corresponding Walsh diagrams for bending CpCoL (L = u donor, 2a” doubly occupied). The different Hi values used for the model ligand’s 1s function are given in each diagram as well as HOMO-LUMO gaps and 2al stabilization energies at the C, minimum energy structures.

destabilized most. Of course not only the energetic position of the donor orbital of L is of importance. As suggested by the standard perturbation expresssion,38 governing the interaction of two MO levels, the overlap also plays a decisive role. The trends here can be tested by changing the exponent of the 1s function of the model ligand L, and parallel effects to changing the orbital electronegativity are caused. The larger the overlap of .the donor orbital with x z (in C,) and z2 (in C 5 J , the more preference toward a bent singlet results. The overlap situtation on the other hand is also influenced by the nature of the metal. The heavier metals (Rh, Ir) should also, from that point of view, be better candidates for a bent singlet structure than their cobalt analogues. We have also performed model calculations for somewhat more realistic donor ligands L with sp3-type donor levels. Examples are CpCo(PH3), CpCo(CH3)-, CpRh(PH,), and CpRh(CH3)- (with an s,p basis for P), and we see the same trends as noted above. The bent geometry is preferred, and the HOMO-LUMO gap depends on the energetic position of the ligand lone pair. We do not present the results here in detail, as they are parameter dependent with respect to the actual numbers, but we note that the methyl group as the better donor leads to stronger stabilization upon bending and to a larger frontier orbital gap (see Table I). Having understood the general picture for ds CpML systems with L as pure u donors, we now turn to more complex ligand types. Actually CpM(PH3)molecules as models for CpM(PR3) systems already take us to a point, where one has to worry about additional effects of acceptor orbitals of PR3 upon structure and possible spin states, as they certainly play a role in real phosphine ligands. (The acceptor levels here may be d AOs on P or u* levels of the P-R bonds of PR3.) ds CpML (L = A Acceptor). Let us first consider a clear-cut and typical back-bonding situation, a linear and “double-faced” a acceptor ligand, the CO molecule. In ~~

~

(38)Hoffmann, R.Acc. Chern. Res. 1971,4, 1.

Figure 5a the interaction diagram between CpCo and CO in C5,, symmetry is given, the two A* orbitals of the CO ligand are seen to interact with 2e ( x z y z ) of CpCo, bringing the half occupied 2e set of the linear structure energetically close to 2al (z2),which is destabilized by the nonbonding donor level n of CO. The l e set of CpCo behaves as in the pure u donor case discussed above, not interacting with the ligand functions. The Walsh diagram for bending to C, is shown in Figure 5b. Again 2e splits into two levels, and 3a’ ( x z ) is destabilized by turning on u antibonding with n and additionally by loss of a interaction with the appropriate a* orbital of CO. The yz member of 2e, 2a”, in this case, does not stay unaffected but is also going up in energy due to its reduced back-bonding to the other **-acceptor level of the CO group. The relevant changes in character for the two frontier levels 3a’ and 2a” are indicated in 16 and 17. As 2a” suffers much less than 3a‘ from its weakened interaction with a* of CO and does not begin to feel antibonding interaction with n, its ascent in energy is much less pronounced than that of 3a‘.

17

8

20

”

The reduced (C,) symmetry finally again leads to some mixing of z2 (and of x 2 - y2, s, x , z ) into 3a‘ and of some xy (and y) character into 2a”, rehybridizing these MO’s as indicated in 16 and 17. It is these two frontier levels, 16 and 17, that allow one to view CpCoCO, CpRhCO, or CpIrCO, like Fe(C0)4 (C2J, as organometallic analogues of CH2,34permitting simple ways of describing complex systems, where such subunits gather to form dimetallacyclopropanes, clusters, etc. Toward single-faced acceptor

(v5-C&)ML 16-Electron Fragments

2 q

1,

\

-1i

-

-12

-

-13

-

-14

-


3e

1-3

.5c

13:

a

Figure 6. Total energy for bending CpCoCO and CpRhCO: solid lines, 28'' doubly occupied; dashed lines, 3a' and 2a'' each singly

occupied.

b

G9 CO ;

-/

by turning on the stabilizing interaction to a r* acceptor orbital, as shown in 18. A s above, an avoided crossing with

GD

C

0

/ - -

lo"

x 2 - y 2 (la') is seen in the Walsh diagram. In contrast to the pure a-donor case, la'' ( x y ) also gets somewhat stabilized by a r* CO component (see 19). In simple terms

2e

2a;

a

I

17C

I

I

15:

I

I

IX

a

Figure 5. (a) Orbital interaction diagram for "linear" CpCoCO (C5"). (b) Walsh diagram for bending C5" CpCoCO to the C,

geometry.

ligands 2a" will cause strong conformational preferences for geometries with a overlap to this rather high-lying donor z2 in CpCoCO is again the metal orbital that, through the stabilization of the 2a' level, carries most of the energetic preference for bending. This time, reduced antibonding to the donor level n of CO is superimposed (39) The computed rotational barrier for a CH2 ligand in a model complex CpRh(CO)(CH,) is greater than 25 kcal/mol: Hofmann, P. unpublished results.

one could say that the bent CpCoCO structure is better than the linear one (on the basis of one-electron MO levels) because there is improved back bonding to the ligand (via y z , z2, and xy in C, as opposed to only y z and x z in C5J. This is clearly reflected by the computed metal charges of CpCoCO in both geometries (C5,,,0.112-; C,, 0.256) as well as in the reduced overlap populations to and within the CO group (C5u,ncwc = 0.843, nc-0 = 1.226; C,, ncWc = 0.908, nc-0 = 1.188). The changes in total energy upon bending for the models CpCoCO and CpRhCO are shown in Figure 6 (solid lines). The bent minima for an occupation pattern (2a"Y (Co, (Y = 142O, stabilization 0.716 eV = 16.5 kcal; Rh, LY = 142O, stabilization 0.759 eV = 17.5 kcal) correspond to a HOMO-LUMO gap of 0.62 (Co) and 0.83 eV (Rh), respectively. The dotted curves correspond to having one electron in 3a' and 2a". If we could compare a pure u donor ligand with a hypothetical one of the same donor strength but with additional acceptor levels available, then the conclusion from above is that the availability of acceptor orbitals should cause a more bent structure and should increase the HOMO-LUMO gap. Model calculations on CpRh(PH,) with and without d orbitals on P in fact show this behavior. This choice is as close as possible to adding just acceptor levels to an unaffected donor-of course, the set of d levels also causes slight changes in the lone-pair


1280 Organometallics, Vol. 2, No. 10, 1983 E(eV1 -9

-

-11

-

-

“ \ \

7

-13

17c

-15

-17

-

150

130

a

-

Figure 8. Total energy for CSu C, transformations of model systems CpCoCN-, CpCoCO, and CpCoNO’ (2a” doubly occupied).

-

also is a better u donor than CO (see Figure 7), NO+ is a weaker one, and this counterbalances and tempers the effects of the reduced or increased acceptor influence. Apparently, however, the changes in the ligand accepting capability dominate the picture. It is in principle conceivable that for very strong acceptor ligands the linear C5, geometry (see Figure 5a) could have the 2al level (z2,antibonding with n) above the 2e set ( x z , yz, bonding to a*). Indeed model calculations for CpCoNO+ and C p C o N S give this sort of picture. The very small energetic separation of 2al and 2e suggests a highspin situation for CBVagain. Bending to a C, structure stabilizes the low-spin systems as found for the 2al below 2e case. The empty level 2al on bending rapidly gets destabilized, changes its character, and becomes 3a‘ with predominant x z contribution, the x z member of the occupied 2e set transforms into 2a’ with dominant z2 participation and descends in energy. In other words the inverted 2al above 2e sequence of a linear structure is immediately reversed by mutual x z / z 2 mixing (an avoided crossing) on bending. The computed reversal of the frontier levels 2a1 and 2e for C,, CpCoNO+ and CpCoNS+ may well be an artifact of the specific method and parametrization we have used. Ab initio MO calculations40 and photoelectron spectroscopic work41on the 18-electron seem to system CpNiNO (C,, symmetry give a 2e above 2al orbital pattern for this nitrosyl compound in its ground state. It should be noted, however, that the PE assignment, based upon ASCF calculations, shows ionization from 2al to occur easier than from 2e. This has been interpreted as a consequence of the inapplicability of Koopmans theorem due to large orbital relaxations (about 7 eV) of the metal-centered levels. It might be though that the ground-state SCF calculation itself does not show the correct energy ordering due to the limited basis set used. Of course for a system like CpNiNO, where 2al and 2e both are fully occupied, the energetic ordering of these two levels is irrelevant with respect to the total extent of dative bonding and back-bonding between CpNi and the NO

-

Figure 7. EH frontier orbital energies for the model ligand systems CN-, CO, and NO+. orbital of PH3, but these changes are very minor. For real ligands, of course, a-donating and a-accepting power are always interrelated. So when we attempt to test in calculations the influence of weak vs. strong acceptor properties (say of CN- vs. CO) or vice versa (say of NO+ vs. CO), we inevitably change both of these electronic features of the ligand. This is revealed by Figure 7, a sketch of the EH orbital energies of CO, CN-, and NO+, showing the degenerate a and a * levels as well as the lone-pair donor orbital n of each group. In all cases of Figure 7 the interatomic distances are taken at a constant value as we are not considering the free ligand molecules or ions but rather some geometry of the groups bound to a metal center. Despite this interrelation of donor and acceptor propensities the trends can be discussed separately as deduced from the above CpCoL and CpCoCO cases. The consequence of a lowered a* acceptor set would be to cause the following on bending: more destabilization of 3a‘ ( x z ) and 2a” ( y z ) due to the greater loss of back-bonding and more stabilization for 2a’ (z2) and la” ( x y ) due to stronger interaction with the acceptor levels. Pushing the acceptor set to higher energy will bring about the opposite effects. In total, a better acceptor ligand L should cause a lowspin ds CpML to exhibit a stronger tendency to be bent but will not lead to a very improved magnitude of the HOMO-LUMO gap. We have tested this on both cobalt and rhodium analogues. Figure 8 shows the comparative features for CpRhCO, CpRhNO+, and CpRhCN-. The bending angle is practically the same, CpRhCO gains about 0.1 eV more in going to the C, singlet arrangement than does CpRhCN-, and bent CpRhNO+ is in turn 0.2 eV more stable (relative to the linear geometry) than CpRhCO. The HOMO-LUMO gaps are 0.83,0.82, and 0.77 eV at 142,140, and 144”, respectively for the CO, CN-, and NO+ cases. The total energy curves for a situation with both 3a’ and 2a” singly occupied show the same trend. CN-, of course,

Hillier, I. H.; Saunders. V. R. Mol. Phys. 1972, 23, 449. (41) Evans, S.;Guest, M. F.; Hillier, I. H.; Orchard, A. F. J.C.S., Faraday Trans. 2, 1974, 70, 417. Burnier, R. C.; Freiser, B. S.Inorg. Chem. 1979,18, 906. -(40)

(q5-C&)ML 16-Electron Fragments


ligand. For a 16-electron fragment like CpCoNO+, however, the 2al above 2e pattern, leaving 2e filled and 2al empty, leads to stronger back-bonding to NO (via xz and y z ) than the 2e above 2al orbital pattern. A rather positively charged metal center results as a consequence. Bending to a C, geometry in the earlier discussed 2e above 2al case of CpCoCO had served to reduce the negative charge a t the metal by strengthening the metal to a* back-donation. Bending of a fragment with the 2al level above the 2e set will reduce the high positive charge, leading comparable bonding situations in both cases for a bent low-spin minimum structure. If we consider both a-donating and a-accepting abilities of ligands L in ds CpML together, we can conclude that the best chance for such a fragment to survive on the singlet energy surface should be offered by good a donors with additional good a-accepting capacity. The structure such species will choose to adopt will inevitably be a bent one. What we have dismissed up to now in analyzing r acceptor ligand effects is the influence of the occupied a levels of such groups. We will, however, single out the consequences of a donor levels of ligands in the next section. ds CpML (L = r Donor). A typical a donor ligand would be a monoatomic group like C1-, carrying a set of three degenerate filled p orbitals (lone pairs). So let us take the hypothetical CpCoCl- as a model to find out what a low-lying set of filled orbitals with a symmetry toward the CpCo fragment will do within our picture. Now according to Figure 9a the linear C,, structure exhibits a large energy separation between the half-filled 2e set, destabilized by antibonding to the two appropriate p-donor orbitals of the ligand, and 2al, which feels the a-antibonding interaction to the filled s and the third p level of the attached group. Figure 9b, the total EH energy change for bending to C, (low-spin model, 2a" in Figure 9a doubly occupied) reveals a preference for a bent geometry again, but the total gain in energy due to orbital energy variations is only rather small (0.18 eV = 4 kcal). In 2al ( z 2 ) ,which is stabilized by bending to a much lesser extent than in our earlier model systems, a antibonding is reduced as before, but it is now replaced by r-antibonding interactions to a p-donor A 0 of C1-. The same sort of antibonding is turned on to the xy member of the C5"l e orbital. This is schematically shown in 20 and 21 and is seen in the Walsh diagram of Figure 9c. x 2 - y 2 as

-11 20'

la"

well is destabilized slightly by a antibonding to the ligand. While the LUMO 3a' ( x z ) of the low-spin system behaves analogously on bending as in the ?r acceptor ligand case, the HOMO 2a" ( y z ) now even reveals a small stabilization in energy because bending reduces the overlap and the antibonding interaction to the y A 0 function of the halide ligand as shown in 22.

Q

-

la'

- -12

/ -

la"

:e

I

173

I

I 15C

I

I

130

a

Figure 9. (a) Interaction diagram for CpCoC1- (&). (b) Total energy for bending CpCoC1- (2a" doubly occupied). (c) Corresponding Walsh diagram.


t

/

EleVl


Y

/ v

l-‘II--x

5

-10

-

-1:

-

of that species, if the singlet is lowest in energy as our analvsis of bent CDML fragments indicates. It would be extremely interesting to see the conclusions of our qualitative analysis being tested by “state of the art” ab initio calculations, and we hope to hereby encourage such studies, which then however would have to involve large enough basis sets, extensive geometry optimizations, and an appropriate amount of ~ o r r e l a t i o n . ~ ~ CpCo(O=NCH,), a da CpML Fragment (L = Single-Faced Acceptor). We close our analysis by inspecting the fragment CpCo(NOCH3),9, which has been identified as an intermediate by Bergman et al. in their study of the NO insertion reaction 8 This case is particularly interesting because not only do we have here a single-faced a-acceptor ligand, the nitrosomethane fragment 23 but also

-

w+ A

-12

NO

le -13

p=2\ c L H,H

?T*

-

because several conceivable different ground-state structures come into question for 9. The CH3N0 ligand may be bound in a 7’ fashion via nitrogen, as found in the X-ray structure of 10, the isolated phosphine adduct of 9, with a possibility of giving four alternative structures of linear or bent geometry (24-27). The +bonding mode is also frequently but not exclusively found for metal-bound acyl groups RCO. 25 and 26 can be derived from the linear

la’

-

-9 c;“ co

I H

G

-

cs

0”N,CH3

0” “CH3

o*,p

-

1

Gs

I

co

-w ‘ 7

/CH3

AI

Figure 10. Orbital interaction diagram for “linear”(C,) CpCo(CH,NO), 24.

Generally we can conclude that a-donating character on L will lead to a rather small energetic preference for a bent structure if a low-spin situation is simulated by placing two electrons in 2a”. The developing HOMO-LUMO gap is not big, and the chance for a bent singlet system is not high. Putting one electron each in 3a’ and 2a” makes the tendency to bend practically vanish. To summarize a t this point, de CpML intermediates with 16 valence electrons especially for the heavier metals may well have singlet ground-state structures and will be bent in particular if the metal bound ligands L are good c donors and, in addition, possess a acceptor capabilities. This may well explain the nonexistence of a high energy barrier due to spin forbiddeness in Bergman’s experiment involving dissocation of CpCo(PPh3), to form CpCo(PPh3) as an intermediate in the substitution process. The thermal equilibrium of CpIr(CO)H2and CpIr(C0) + Hz was well as the other observations cited in the beginning could become understandable then as well. We would like to point out here that the oxidative addition of H2 or a C-H bond to a bent singlet da CpML is an allowed ground-state process. This is easily derived from a Woodward-Hoffmann type correlation diagram (orbital or state correlation) for this reaction. In contrast to the isolobal carbene case, where the symmetric least motion insertion into the Hz bond is forbidden, a d8 CpML singlet system with a bent Cp-M-L arrangement has the antisymmetric valence level (2a”) occupied and the a-type level (3a’) empty, leading to the necessary HOMO-LUMO interactions for the least motion insertion reaction. If d8 CpML 16-electron fragments are generated in an initial triplet state in photochemical experiments, intersystem crossing, which becomes possible for such heavyatom molecules, can nevertheless lead to a bent singlet intermediate and to subsequent ground-state chemistry

G s co

I

oqNh3

=

CO

‘N - - - [ H ~ *7

%

I

geometry 24 by bending within the Co-CH3N0 plane toward either the oxygen or the methyl side, and 27 results from bending 24 perpendicular to the Co-CH3N0 plane or, alternatively, from a 90° rotation around the Co-N bond in 25 or 26. Figure 10 presents the computed interaction diagram between a CpCo fragment and the CH3N0 ligand for the geometric arrangement as in linear 24. A t the ligand, two frontier orbitals are relevant, the donor orbital n,localized more at nitrogen with some antibonding p contribution from oxygen and the low-lying a*Noacceptor level of the N=O double bond. The HOMO of 24 derives from lal ( z 2 ) of CpCo, which is destabilized by n of the ligand (again the high-lying 2al orbital of CpCo mixes into this level in a bonding way). The xy and x 2 y2 levels ( l e of CpCo) of the metal are practically unaffected by the ligand. The x z member of 2e of the metal fragment does not feel much ligand interaction either and stays put. The reader will so far have recognized exactly our old pure u donor case discussed earlier. The important additional interaction for 24 stems from T*NO of CHBNO and the yz member of 2e. The bonding combination of these two levels, due to the low energy of a*No, is found close above xy and x 2 - y 2 of the metal, far below z2, This level is occupied by two electrons, and thus the picture corresponds to the 2al above 2e situation that we have (42) Veillard, A,; Dedieu, A,, some ab initio work on CpRhCO, submitted for publication in Theor. Chirn. Acta.

( $ - C a 5 ) M L 16-Electron Fragments


-1c

-11

/ la'

-:2

I 1

I I

-13

I Ila'

-4c

-

160

180

160

a

25

b

Figure 12. Frontier level Walsh diagram for going from 24 to 27. The frontier levels alone, as shown in the figure, still would induce some gain in total energy on bending to 27. The overall loss in energy for going from 24 to 27 is mainly due to a series of lower lying levels not included in the diagram.

Figure 11. (a) Walsh diagram for transforming 24 into 25 (left) and 26 (right) and (b) corresponding total energy behavior.

described for strong double-faced a accceptors before. A HOMO-LUMO gap of about 0.75 eV is found already for such a linear structure 24 of CpCo(CH3NO). From what we have learned about the simplified model systems in the earlier sections, it is clear that 24 will have an electronic tendency to also adopt a bent geometry, either 25,26, or 27. Bending in any case will stabilize the z2-derived HOMO by reducing the antibonding to n, and this is seen in Figure l l a , the Walsh diagram for transforming 24 into 25 (left) or 26 (right). For these two geometries the x z level (4a') of the linear form is going up in energy as in all other cases before. The highest empty level shown in Figure lla, 3a" ( y z ) of CpCo antibonding to T*NO, is going down slightly for the same reason that causes the A*NO + yz (2a") MO to go up and counteract the stabilization of z2. The x 2 - y2-based orbital (2a') gets somewhat destabilized while xy starts to interact with K * ~ Oand gains in energy. According to Figure l l b both bent structures 25 (a = 147') and 26 (a = 150") are energetically better than 24. 25 is preferred by about 0.17 eV over 26, and this slight preference is mainly a result of steric factors and of the somewhat differing overlap situation of z 2 with n for the two ligand orientations in 25 and 26, leading to a lower lying z2-type level in 25. The HOMO-LUMO gap for the

two bent geometries, where the two lowest empty levels derived from x z and yz at the metal become nearly degenerate, increases to 1.5 eV. This gap, within the EH framework of theory, ensures a singlet ground state for such bent CpCo(CH,NO) 16-electron intermediates and indicates that reactions via such species can take place without problems on the singlet energy surface. Transforming linear CpCo(CH3NO) (24) to the bent geometry 27 is found to be unfavorable. From the MOs of 24, displayed schematically in Figure 10, the reader can easily see that bending perpendicular to the Co-ligand plane will leave the LUMO (xz) unaffected and will reduce the a interaction between yz and a*No,replacing it by 0 antibonding of yz to n. Bending of 24 toward 27 also induces a interaction of z2 with a*N0.This causes the HOMO of 24 not to descend in energy because now the stabilization due to reduced antibonding to n of the ligand is undone by the mixing of into z2 in an antibonding fashion. It is only the bonding combination of z2 and which is stabilized upon going from 24 toward 27. Figure 12 displays the energy level changes for this geometric distortion. In contrast to the stable bent structures 25 or 26 bent 27 has a smaller gap between HOMO and LUMO than even linear 24. This is also easily derived if one goes to 27 by rotation around the Co-N bond of either 25 or 26. To follow this can be left to the reader. So far we have kept the CH3N0 ligand bound to Co in a manner. In fact it is well-known that in various stable and isolable acyl complexes of transition metals electronic unsaturation at the metal center is circumvented by coordinating acyl groups in a q2 was" (see 28). This has been


1284 Organometallics, Vol. 2, No. 10, 1983 Table 11. Extended Huckel Parameters

CO 3d 4s 4p Rh 4d 5s 5p H b 1s P ~ s 3p 3d Cl 3s 3p

-12.11 -8.54 -4.76 -13.6 -8.77 -5.20 -13.6 -18.6 -14.0 -7.0 -30.0 -15.0

5.55 2.0 2.0 4.29 2.135 2.10 1.3 1.6 1.6 1.4 2.033 2.033

2.1

0.56786

0.60568

1.97

0.58070

0.56850

a Coefficients in the double-2 expansion. Both Hii value and 1 s exponent were varied to mimic a change in 0-donor ability. For explanation see text.

discussed also as one possibility to explain the stereochemical behavior of 16-electron intermediates of the CpML(acy1) type like 29.& For CpCo(CH,NO) this might suggest a structure like 30. Indeed our calculations reveal 30 (with an idealized, symmetric45ligand arrangement) to be of comparable energy to the bent p’ species 25 and to also have a large HOMO-LUMO gap of about 1.2 eV. The HOMO now is the antibonding combination of xy and n*NO, shown in 31. The LUMO, 32, results from antibonding of “*NO with yz. The oxygen lone pair fills the

-

31

CH3

CH3 32 -

valence shell of the cobalt atom, and 30 in our calculations finds its minimum energy with a “planar” ligand arrangement as generally found for CpML, 18-electron complexes, e.g., CpCo(CO),, CpCo(PR&, etc. We hesitate, however, to assign a relative energy ordering to the two stable minimum structures of CpCo(CH3NO) (9), because the structure would require optimization of at least some of the bond lengths. This cannot be done reliably within the E H model. In any case, an intermediate da CpM(CH,NO) should be a system with a singlet ground state. Conclusions

Qualitative MO theory, supported by EH calculations, (43)Fachinetti, G.;Floriani, C. J. Organomet. Chem. 1974, 71, C5. Fachinetti, G.; Floriani, C.; Stoeckli-Evans, H. J. Chem. SOC.,Dalton Trans. 1977, 2297. Fachinetti, G.;Floriani, C.; Marchetti, F.; Merlino, S. J. Chem. SOC.,Chem. Commun. 1976, 522. Fachinetti, G.; Fochi, G.; Floriani, C. J. Chem. SOC.,Dalton Trans. 1977, 1946. Fagan, P.J.; Manriquez, J. M.; Marks, T. J.; Day, V. W.; Vollmer, S. H.; Day, C. S. J. Am. Chem. SOC.1980, 102, 5393. Fagan, P. J.; Manriquez, J. M.; Vollmer, S. H.; Day, C. S.; Marks, T. J. Ibid. 1981, 183, 2206. Franke, U.;Weiss, E. J. Organomet. Chem. 1979, 165, 329. (44)Brunner, H.; Vogt, H. Angew. Chem. 1981, 93, 409. Angew. Chem., I n t . Ed. Engl. 1981,20,405.Brunner, H.; Vogt,H. Chem. Ber. 1981, 114, 2186 and references therein. (45)Equal bond distances from Co to N and 0, angle 0-N-C being 1200.

shows that 16-electron intermediates ds CpML with appropriate ligands can adopt bent (C,)singlet structures in their electronic ground state. Especially good candidates for bent singlet structures ares systems with good a-donating, n-acceptor ligands and with the heavier metal atoms. Many reactions, where such species have been invoked as essential intermediates formed by thermal or photochemical ligand dissociations or by other routes, can thus proceed on the singlet energy surface without any spin-induced barriers. Acknowledgment. We are grateful to the Deutsche Forschungsgemeinschaft and to the Fonds der Chemischen Industrie for generous support of this work. The cooperation of the staff of the Regionales Rechenzentrum der Universitat Erlangen is also gratefully acknowledged. We thank Prof. Roald Hoffmann and Dr. B. E. R. Schilling for helpful discussions and Prof. A. Veillard and Dr. A. Dedieu for providing results of their work prior to publication. Appendix

The parameters used in the extended Huckel calculations are given in Table 11. The modified WolfsbergHelmholz formula was used throughout the calculation^.^^ Hii values for cobalt as well as wave functions were taken from earlier work:’ Hii values for Rh were obtained from charge-iterative calculations on CpRh(CO),, and the Rh orbital exponents are those used in previous The main-group parameters are standard ones. To model weak and strong donors the 1s Hii of hydrogen (the L model ligand) was modified as indicated in the main body of the paper. P was considered with and without d functions. Idealized C-C (1.41 A) and C-H distances (1.09 A) were taken for the Cp groups in all cases. The Co-Ccp and Rh-Ccpdistances were 2.088 and 2.25 A, respectively. The Co-CO distance was taken as 1.728 A, and the same value was employed for Co-CN and Co-NO to compare these ligand types rationally. For the Rh analogues these distances were set at 1.83 A. On similar grounds the c-0, C-N, and N-0 distances in all the diatomic ligands were kept constant at 1.135 A. The other distances used were Co-H = 1.556 A, Co-P = 2.145 A, Co-CH3 = 2.0 A, Co-C1 = 2.3 A, Rh-H = 1.72 A, Rh-P = 2.30 A, Rh-CH3 = 2.1 A, and Rh-C1 = 2.37 A. For the nitrosomethane complex 9 the bond distances were adapted from its PPh,-containing analogue reported in ref 23. Registry NO.CpCo(PHJ, 86803-00-1; CpRh(PH3),86803-01-2; CpCo(CH,)-, 86803-02-3; CpRh(CH3)-, 86803-03-4; CPCOCO, 73740-50-8;CpRhCO, 86803-04-5;CpCoCN-, 86803-05-6;CPCONO+, 86803-06-7;CpCoCl-, 86803-07-8;CpCo(CH,NO), 86803-08-9. (46) Ammeter, J. H.; Burgi, H.-B.; Thibeault, J. C.; Hoffmann, R. J. Am. Chem. SOC.1978,100, 3686. (47)Albright, T.A.;Hofmann, P.; Hoffmann, R. J. Am. Chem. SOC. 1977, 99,7546. (48) Summerville, R. H.; Hoffmann, R. J. Am. Chem. SOC.1976, 98, 7240.

Electronic and Geometric Features of (q5-C5H5 ... - ACS Publications

Recommend Documents